Implementers of RF-indentification (RFID) readers have sometimes struggled with meeting read rate requirements due to the difficulty of isolating different factors-that can affect performance. These include the material to which the tags are attached, the orientation of the tags, power constraints, operating frequency variations, and other factors. The large number of design parameters for evaluating RFID read-rate performance translates into a large work space for physical experiments, not always practical in an actual design environment. Fortunately, a relatively new approach to analyzing RFID read-rate performance involves the use of electromagnetic (EM) software simulation tools to evaluate multiple system configurations rapidly while obtaining more information from each analysis than is possible with physical testing. Using this approach engineers can quickly identify-the root cause of RFID read rate problems and evaluate prospective solutions in a fraction of the time required with experimental methods.

Typically, read-rate problems are usually detected after the RFID equipment has been installed in a manufacturing, distribution, or retail facility. At this point, the implementers typically try to solve the problems by trying different tags, readers, antennas, moving equipment to different locations, etc. Unfortunately, after the equipment has been installed, even minor changes can be very disruptive to the facility where the equipment is installed. For these reasons, many applications have found it impossible to achieve anywhere close to the 95-percent read-rate level that is a common goal.

RFID equipment manufacturers and systems integrators are just beginning to turn to computer software simulation tools as a potential method for identifying, understanding, and solving RFID read rate problems inexpensively before the equipment is installed. The basic idea is to simulate the performance of the reader and tag in a simulated environment that duplicates the conditions in the facility where they will be installed. Predicting the return rate and loss of RFID tags and readers under real-world operating conditions makes it possible for engineers to evaluate different design parameters long before installation in an environment where various alternatives can easily and quickly be evaluated without investing in new hardware or disrupting production.

It's important to note that the complexity of RFID read-rate performance factors sometimes requires that simulation be performed at several levels, such as the tag, the product package, the cabinet, and the surrounding environment. The first step is normally to model the tag in isolation using a modeling tool that can represent arbitrarily shaped antenna structures. As an example, a 900-MHz RFID tag model (Fig. 1) was excited by means of a wire feed placed between the two arms of the dipole. The model was then solved to compute the response of the tag with no dielectric support. The return loss of the antenna was calculated to be 17 dB at 898 MHz (Fig. 2). At this frequency, the dipole type radiation pattern was accurately predicted, presenting a directivity of 1.9 dBi. A higher-order resonance was also captured in the response at around 2.6 GHz.

The next step in the simulation and evaluation process was to mount the tag on a sheet of corrugated cardboard in order to introduce the effects of packaging on tag performance. Two planar carton sheets of 200 X 100 mm and 0.5 mm thick were situated 3 mm from one another. The upper sheet was positioned such that the antenna element sat on its surface (Fig. 3). Then, the spacing between the sheets was filled with the corrugations which consisted of a semicylindrical tube contiguously copied a number of times as necessary to cover the whole length of the two cardboard sheets. The material chosen for the cardboard was paper, having a dielectric constant of 2.5. When the antenna is mounted on the cardboard, computer EM simulation shows an approximate 55-MHz shift in frequency for the return-loss performance, to 843 MHz. The return loss also improved by about 2 dB to 19 dB. The radiation pattern remained consistent with the previous result, as did the directivity at around 1.9 dBi.

The corrugated cardboard was then removed and, the antenna mounted on a 4-mm-thick plastic board of the same size. The dielectric constant of the plastic material was 4.7. Once again, the results show another shift in frequency. The antenna is now yields return loss of about 24 dB at 683 MHz. The radiation pattern and directivity at this frequency were shown to be consistent with the previous two cases (Fig. 4).

This capability to study the effects of mounting an RFID tag on different materials is very important as it allows a designer to tune the antenna to resonate at the desired frequency, taking into consideration the dielectric loss of other materials in its installed environment. For the antenna shown in this example, the first option would be to make the dipole shorter. Through simulation, the resonant frequency was designed back to around 900 MHz. However, further design would be needed to improve the return loss.

To add some sophistication to these computer evaluations, the RFID tag was simulated as being placed on items sitting on shelves, using a "smart cabinet" with a reader antenna mounted behind a backplane (Fig. 5). This is an auto inventory system configured so that any item deposited in or withdrawn from the cabinet will be automatically detected by the reader antenna and recorded into a log.

The goal of this simulation is to determine the level of coupling between the reader and the tags when items are placed in the cabinet. Understanding the coupling levels between readers and tags is a crucial part of RFID system design. Through the use of the built-in visualization tools in modern simulation software, it is possible for a system integrator to gain added insight into the characteristics of the cabinet allowing them to determine the most effective location for the reader as well as the optimum number and location of items in the cabinet.

The cabinet in this example has three shelves with a single, arbitrarily placed tagged box on each shelf. As per the previous example, the tags and the reader are in the 900-MHz UHF band. To simplify the model, and in order to reduce the simulation time, the cabinet walls and shelves are constructed from thin metal sheets. The boxes are defined as thin film dielectric materials.

The fastest approach to calculate the coupling between the reader and the tags is to drive the reader (port 1) with an input signal and output the transmitted signals arriving at the tags (ports 2, 3 and 4). This yields the S-parameters S21, S31, and S41.

The blue line in Fig. 6 is the input signal in Port1 (reader antenna). The red plot is the coupling to the port attached to the tag sitting on the middle shelf. We would expect this coupling to be the strongest as there is no obstacle between this tag and the reader. The other two plots (green and cyan) are the tags situated on the top and on the bottom shelf. Translated into dB, these results show a path loss of approximately 0.1 dB for the central tag and approximately ?11 dB for the tags on the bottom and top shelves.

With values such as these, it could be assumed that there is sufficient coupling between the tags and the reader to ensure correct identification of the items. Should a cabinet have more shelves or have more items inserted, then this response would differ and the communication to the extremities could be hampered. Then a new reader location or another type of reader, such as a loop antenna, could be considered.

For these simulation cases the RFID tags are passive (not powered); they communicate with a reader by re radiating a portion of the signal emitted by the reader. The product that is being marked by the tag will disturb the re radiation from the tag. The illustration shown below represents the packaging for a bicycle with the tag on the outside of the box (Fig. 7).

In the example shown here, the bicycle has metallic components that pick up the reader field and radiate it in a manner that interferes with the tag (Fig. 8). The components create nulls in the backscatter pattern that prevents tags located in the null area from receiving any signal at all (Fig. 9). In this case, understanding the cause made it possible to create a more uniform radiation pattern.

Computer-aided EM simulation has the potential to improve the reliability of RFID systems dramatically by making it possible to evaluate a wide range of potential configurations prior to installation in order to evaluate and optimize system performance inexpensively without disrupting operations. The latest advance in RFID simulation enables a series of simulations to be run automatically while varying one or more design parameters over a user-specified range. This feature speeds up the design process by making it possible to, for example, quickly consider a wide range of locations and determine the optimal reader placement.